Abstract
Since computerized tomography (CT) images have general structures consisting of an approximately elliptical region containing almost all of the image energy and are bounded by a region in which the image intensity is essential zero, the property must be introduced into the source model for performing transform coding analysis. For 3-D CT image data sets, in order to take advantage of the time-domain correlation between contiguous images, the difference between consecutive images is encoded instead of the original one. In this paper, energy compaction capabilities of DCT and DHT using 2-D source model (for spatial dependence) and difference model (for time dependence) with boundary constraints, both separable and non-separable, are explored. Experiments are performed to verify that DHT is suitable to be used in transform coding for CT data.